Quick Answer
New hires who follow a structured onboarding plan reach full productivity far faster than those left to figure it out. AI is what makes 'structured' cheap enough to do for every single hire.
What you get
- Generate a role-specific 30-60-90 plan, welcome doc, and first-week schedule in ~30 minutes per hire
- Cut the manager's per-hire onboarding-doc prep from ~4 hours to ~30 minutes
- Deflect the majority of repetitive week-one questions to a self-serve AI answer layer
- Run the core system for $56/mo, or $350/mo with async video and formal training tracking added
Step-by-Step Workflow
5 steps · 5 hours to set up · ~30 min per new hire ongoing
Workflow at a glance
5 steps · 5 hours setup
Build template
NotionGenerate plan
ClaudeNotionAnswer layer
ClaudeAsync video
LoomCheck + iterate
ClaudeBuild template
Generate plan
Answer layer
Async video
Check + iterate
- 1
Build one role-flexible onboarding template
Create a single Notion template with the sections every onboarding needs: a welcome and what-we-do overview, a first-week schedule, a 30-60-90 plan with concrete milestones, the tools-and-access checklist, a glossary of internal terms and acronyms, and a 'who to ask about what' map.
The milestones are the part most onboarding skips and the part that matters most: 'by day 30, ship one small thing to production' beats 'get familiar with the codebase' every time. Build the template once so every future hire starts from a proven structure.
Notion - the interface you'll work in for this step. Screenshot of the tool's own UI, not our results. 2 hrsOutput: A reusable onboarding template with concrete 30-60-90 milestonesTools: NotionTip: Write the 90-day milestone first, then work backward to 60 and 30. Starting from the endpoint forces the plan to build toward a real outcome instead of a list of orientation activities.
- 2
Turn the template into a role-specific plan with AI
For each new hire, feed Claude the role, the team, the scorecard or job description, and your template structure. It fills in a plan tailored to that role: the specific things a marketer versus an engineer versus a CS rep should learn and do in each window, matched to your template's milestones.
This is the step that used to cost the manager 3-4 hours per hire and now costs about 30 minutes including the human edit. Always edit the output - the AI gets the structure right but you know which project is actually the right first win.
Claude - the interface you'll work in for this step. Screenshot of the tool's own UI, not our results. 30 min per hireOutput: A tailored 30-60-90 plan, welcome doc, and first-week schedule per new hireTools: Claude, Notion - 3
Stand up a self-serve answer layer
New hires ask the same 40 questions - how do I request time off, where is the brand kit, who approves expenses, what does this acronym mean. Point an AI over your onboarding docs so they can ask in plain language and get an answer without interrupting a teammate.
The lightweight version is a Claude Project loaded with your policies and docs that the new hire can query. The heavier version is a dedicated tool like Guru sitting over your knowledge base. Either way, the goal is the same: deflect the repetitive questions so human time goes to the questions that actually need a human.
1 hrOutput: A plain-language Q&A layer that deflects the majority of repetitive week-one questionsTools: ClaudeTip: Log the questions the answer layer could not handle for the first few hires. Each gap is a doc you are missing; fixing them makes the next onboarding smoother for free.
- 4
Record the walkthroughs once, reuse forever (optional)
Some things are faster shown than written: the deploy process, the CRM workflow, the way your team runs standup. Record these once with Loom and drop them into the onboarding hub. A new hire can watch at their pace and rewatch, and the manager never gives the same live walkthrough twice.
Use AI to draft the transcript-based summary and chapter markers so each video is searchable and skimmable, not a 20-minute block someone has to sit through linearly.
Loom - the interface you'll work in for this step. Screenshot of the tool's own UI, not our results. 2-3 hrs to record the core set onceOutput: A reusable library of async walkthroughs that replace repeated live demosTools: Loom - 5
Check comprehension and close the loop
Onboarding without a feedback loop silently decays. Use Claude to generate a few light comprehension checks tied to the plan's milestones - not a test, a way to surface what did not land. Pair that with a genuine human check-in at day 7, 30, and 60.
After each hire, ask one question: what did they struggle to find or understand? Feed the answer back into the template and the answer layer. The system should be measurably better at hire number ten than at hire number one.
Claude - the interface you'll work in for this step. Screenshot of the tool's own UI, not our results. 30 min per hire + template updatesOutput: Milestone comprehension checks and a template that improves with every hireTools: Claude
Onboarding is expensive to do well and disastrous to do badly, and most teams do it badly for the same reason: building a real, role-specific plan for every hire takes hours the manager does not have, so the new person gets a generic doc, a firehose first day, and a month of quietly not knowing what to do. AI removes that cost. A People-ops owner builds the system once - a template, a set of prompts, and a self-serve answer layer - and then each new hire gets a plan tailored to their role in about thirty minutes. This is not about replacing the human parts of onboarding, which are the parts that make someone feel they belong. It is about deleting the manual document assembly so managers spend their onboarding hours on the relationship and the context, not on formatting a checklist.
What AI onboarding should and should not touch
AI is excellent at the structuring and content layer of onboarding: turning a role and a scorecard into a 30-60-90 plan, drafting the welcome doc and glossary, generating a first-week schedule, writing comprehension checks, and answering repetitive 'where do I find X' questions from your existing docs. That is where the manager hours and the new-hire confusion both live. What AI should not own is the human side - the manager 1:1s, the intros, the culture and context that only a person can transmit, and any judgment call about the new hire's ramp. A good rule: AI handles the information, humans handle the belonging. Onboarding that gets that split wrong feels cold and automated; onboarding that gets it right feels both organized and warm.
Stack cost breakdown
Public list prices as of July 2026. Optional tools are marked in the notes.
| Tool | Plan | Monthly cost | Notes |
|---|---|---|---|
| Notion | Plus | $36/mo | Required. Onboarding hub, 30-60-90 templates, checklists. $12/seat/mo x 3 People/ops editor seats. |
| Claude | Pro | $20/mo | Required. Generates role-specific plans, welcome docs, quizzes, and the answer layer's content. |
| Loom | Business | $45/mo | Optional. Async walkthrough videos. $15/creator/mo x 3 creators. |
| Trainual | Small team | $249/mo | Optional. Structured training with completion tracking; flat up to ~10 seats, quote-based above. |
| Total | $56 - $350/mo($56 required, $350 with optional tools) | ||
Email me this stack as a checklist
Every tool, the plan to pick, and the monthly cost - in your inbox.
Real usage
What people actually run
No usage reports yet - be the first to share what you run. Tell us your real stack, your actual monthly cost, and any tools you swapped.
Prompts to copy
The plan-generation prompt is the load-bearing one - it is what turns a generic template into a role-specific plan in minutes. Edit the output; AI gets the structure right but you know the right first project.
Role-specific 30-60-90 prompt
Generate a 30-60-90 day onboarding plan for a new {role title} joining the {team} team at {company, one line}.
Context:
- What this role is accountable for: {2-3 bullets}
- Tools they will use: {list}
- The first real project I want them to own: {the actual first win}
Use this structure for each window (30/60/90): Learn (what to understand), Do (concrete tasks with an owner and a checkpoint), and Milestone (one measurable outcome). Make the day-30 milestone something they can genuinely ship. Be specific to this role - no generic 'get familiar with the team' filler.Note: The 'first real project I want them to own' line is what makes the plan useful. Vague input produces a vague plan; a real first win produces a plan that builds toward it.
Welcome doc + glossary prompt
Write a warm, concise welcome doc for a new {role} at {company}. Include: a 3-sentence 'what we do and why it matters' overview, their first-week schedule (I will fill in the meeting times), and a glossary of the 10-15 internal terms or acronyms someone in this role will hear in week one - here are ours: {paste terms}. Tone: friendly and human, not corporate. Short paragraphs.Note: The glossary quietly removes a huge amount of week-one confusion. New hires nod along to acronyms they do not know for weeks; writing them down once fixes it.
Adjust for Your Situation
If you are a small startup without a People team
Skip Notion Plus seats beyond one and skip Trainual entirely. A single Claude Pro subscription plus a free Notion or Google Doc hub runs the whole thing for ~$20/mo. The founder or hiring manager owns the template. You lose formal tracking but keep the core win: every hire gets a real, role-specific plan instead of a firehose first day.
If you hire in a regulated industry
Add Trainual or a formal LMS for the parts that need documented completion - compliance training, security policy acknowledgment, certifications. Keep the AI-generated role plan for the productivity side, but do not rely on AI or an honor-system checkbox where you need an auditable record that a specific person completed specific training.
If you are onboarding fully remote hires
Lean harder on the async video layer - Loom becomes closer to required than optional, because a remote hire cannot absorb context by osmosis. Front-load the walkthroughs, over-invest in the 'who to ask about what' map, and schedule more frequent human check-ins in the first two weeks to replace the ambient connection an office provides.
Swap options
Drop-in substitutions if a tool does not fit your budget or stack. These trade cost or effort for the recommended setup.
| Swap out | Use instead | When |
|---|---|---|
| Notion | Guru or a Google Sites hub | You want an AI 'ask anything' layer over existing docs rather than building the hub in Notion |
| Claude | ChatGPT or Gemini | Your team already works in one of them; the plan-generation prompts are identical |
| Trainual | Notion databases with a completion checkbox | You do not need formal compliance tracking and want to avoid the $249/mo |
| Automate provisioning and paperwork | Rippling or Gusto for HRIS-driven app access and e-sign | Provisioning is the bottleneck rather than knowledge transfer |
Common Pitfalls
- Letting AI replace the human parts of onboarding. A perfectly organized hub does not make someone feel they belong - the manager 1:1s and intros do. Automate the information, never the belonging.
- Shipping the AI-generated plan without editing it. The AI nails structure but does not know which first project is actually the right win for this hire on this team. Always do the human pass.
- Building the template and never updating it. The whole value is that the system improves with each hire; if you never feed back what confused people, it decays into the same generic doc you were trying to escape.
- Front-loading everything into day one. A firehose first day is the classic failure. The 30-60-90 structure exists to pace the information; do not undo it by dumping all fifteen docs on the new hire before lunch on Monday.
Frequently Asked Questions
Does AI onboarding feel cold or impersonal to new hires?
Can I run this without a dedicated People or HR team?
How is this different from just buying an HRIS like Rippling or Gusto?
Do I need Trainual, or is Notion enough?
How do I keep the AI answer layer from giving wrong answers?
How we built this workflow
This system reflects structured onboarding built on 2026 AI tooling for small and mid-size teams. Prices are 2026 monthly rates verified in July 2026: Notion Plus at $12/seat, Claude Pro at $20, Loom Business at $15/creator, and Trainual at a flat ~$249/mo for small teams (quote-based above ~10 seats). The core $56/mo figure assumes three People/ops editor seats plus one Claude Pro; the small-startup variant runs near $20/mo. The productivity claims assume the human check-ins are kept, not replaced.
Last updated July 7, 2026; prices verified at publication.